The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown documentation and instructions. No executable scripts, shell commands, or network-active code were found.- [INDIRECT_PROMPT_INJECTION]: The skill identifies that data sources like scraped text, reviews, comments, and emails are untrusted inputs. It includes explicit defensive instructions to the agent to treat these as evidence only and to disregard any instructions embedded within that data.- [CREDENTIALS_UNSAFE]: The skill contract and connector documentation (references/connectors.md) explicitly state that API keys must be kept out of skills and source control, following security best practices.- [COMMAND_EXECUTION]: The skill uses a human-in-the-loop model, explicitly stating that persistent file changes, account mutations, or external side effects (like sending messages or spending budget) require direct user authorization.- [EXTERNAL_DOWNLOADS]: The skill references documentation and pricing from Meta/Facebook for WhatsApp services. These are well-known technology service providers and the references are purely informational.

sms